The Iron Streets of Pompeii Author(S): Eric E

Home , Garum

The Iron Streets of Pompeii Author(s): Eric E. Poehler, Juliana van Roggen and Benjamin M. Crowther Source: American Journal of Archaeology , Vol. 123, No. 2 (April 2019), pp. 237-262 Published by: Archaeological Institute of America Stable URL: https://www.jstor.org/stable/10.3764/aja.123.2.0237

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Archaeology

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms The Iron Streets of Pompeii , ,

Includes Supplementary Content Online

In July 2014, we conducted a survey of Pompeii’s street network to document traces of iron that were observed on the stone-paved streets, which resulted in the identification of 434 instances of solid iron and iron staining among the paving stones. This paper de- scribes the iron deposits, categorizing them into six observable types, and argues that, in the final days and weeks before the eruption of Mount Vesuvius in 79 C.E., Pompeians werein addition to using solid iron wedgespouring molten iron and iron slag onto their streets as a method of emergency repair. Before discussing the evidence available for how the melting, transporting, and depositing of these ferric materials might have been accomplished, we address whether the Romans had the technical ability to achieve sufficiently high temperatures to melt iron, finding much evidence to affirm the claim that they did. Finally, we consider why Pompeians undertook such measures to repair their streets. Recent research on the costs of paving stone streets in terms of time, money, and opportunity provides the economic context for this novel repair process and shows the use of iron and iron slag to have been an expedient alternative.1

By the late first century C.E., within only decades of the Roman conquest of Britain, the iron-rich region of the Weald was already producing more than 550 tons of wrought iron annually,2 making it one of the largest iron production sites within the Roman empire.3 Extracting the raw iron from the stone ore that encased it required reducing the ore to a slurry so that a viscous sponge-like bloom of iron collected within the furnace, while the remaining materials were liquefied and drawn off as slag. Later, the bloom was reheated and hammered to remove remaining impurities and give shape to the metal. The bloomery process, however, was surprisingly inefficient and a significant percentage of the iron, from 40% to 70%, remained in the slag,

1 e authors would like to extend their thanks to Massimo Osanna and the superintendency and custodians of Pompeii for providing access to conduct this research. e Pom- peii Quadriporticus Project and Pompeii Archaeological Research Project: Porta Stabia also provided years of infrastructural support. We are also grateful to e Five Colleges, Inc., for early funding through their Digital Humanities Fellows program and to the anon- ymous reviewers for the AJA who helped us improve this paper. Special thanks belong to American Journal of Archaeology Mark Robinson, who walked Pompeii’s streets with us to examine many iron deposits and Volume 123, Number 2 whose challenges and encouragements were of the greatest value. e Supplementary Ta- ble for this article can be found online at hps://works.bepress.com/eric-poehler/107/. April 2019 Figures are the authors’ own unless otherwise noted. Pages 237–62 2 Cleere 1976; Cleere and Crossley 1985; Rackham 2001, 40–1. DOI: 10.3764/aja.123.2.0237 3 See Pleiner (2000, 41–7) for an overview. Craddock (2008, 108) reports the annual iron output of the Roman Empire was 82,500 tons, but see Sim (2011, 22–3) for other www.ajaonline.org estimates.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . which the ironworkers subsequently deposited in enormous waste piles.4 In some cases, this waste material was recycled or repurposed for building material; the largest example of reuse of slag was as paving for long stretches of the London-Lewes Way in Sussex. Averaging 4.5 m to 5.5 m wide and as much as 0.35 m deep, the slag covered more than 30 km of roadway, an amount that Straker and Margary calculate to be no less than 35,000 tons.5 Over decades of exposure and use, one stretch of the roadway, near Holtye, became oxidized and fused into a single, seamless, ferrous mass (fig. 1). The deep ruts found worn into this surface were likely created by the iron rims of wagon wheels made of the very metal separated from the slag paving. This is England’s great iron road, a curious artifact and testament to the scale of industry in the imperial age. Only a few other routes paved with slag are known from antiquity. These include a second road in Britain (in the Forest of Dean), this one connecting a mine and production facilities.6 Elsewhere, Roman cities such as Cardiff (Wales) and Rouen (France) paved some of their urban streets in slag. In fact, at Rouen the practice was widespread; multiple streets were covered with . . London-Lewes Way, a Roman road near Holtye, Sussex, slag, including one street that Romans paved in this paved in iron slag (Margary 1965, frontispiece). fashion multiple times.7 While slag paving is rare, deep ruts are a famous aspect of Roman roads everywhere. Investigations at Pompeii have shown that particularly erroneous, belief that Romans could not achieve tem- high volumes of traffic concentrated in narrow streets peratures necessary to melt iron.9 To explore the use of could wear down even a stone-paved surface in only a iron in Pompeian streets, we first describe our method few decades. One option for repair, complete repav- of study and our classification of the more than 400 ing in stone, was a difficult and expensive endeavor instances of iron used in solid and in liquid forms in that might block important through routes in a city Pompeii’s streets. Next, we address the technical fea- for months.8 Research by the authors has revealed that sibility of such a process and attempt to reconstruct the Pompeians devised another option that was inge- where and how it might have been accomplished. In nious and unconventional: after heating iron or iron- the final section, we attempt to answer to why such an rich slag to a molten state, they poured out hundreds unusual procedure was undertaken. of individual repairs onto, into, and below the paving The primary unit of our analysis is the individual de- stones of the city’s most important streets. This claim posit of iron or ferric material. We have divided these is controversial not only for the novelty of such a pro- deposits into six types. The online supplementary cedure but also because of the commonly held, but table Description of Iron Deposits (at https://works. bepress.com/eric-poehler/107/) presents these data.10

4 Tylecote 1986, 175–76; Craddock 2003, 233; Saredo Paro- di 2013, 18–22; Pérez Macias et al. 2014, 15–9. 9 Specialists in metallurgy have long been disabused of this 5 Straker and Margary 1938, 58. On calculating slag heaps belief (e.g., Read 1934, 545–46), but modern reference works more recently, see Humphris and Carey 2016. continue to repeat the idea (e.g., Humphrey et al. 1998, 218). 6 Pleiner 2000, 45, 267. Davies (1935, 90) notes slag-paved Even in 2012, the Oxford Classical Dictionary entry on metal- roads in Normandy near Mayenne and Tourouvre (Mézières). lurgy states that iron could not be melted until the 19th century 7 Deglatigny 1931, 177, 185, 187–90, 195, 208–9; Davies (OCD 2012, 939). 1935, 90–1; Schubert and Ingen Housz 1957, 41 n. 2. 10 is article’s online supplementary content is hosted by the 8 Poehler and Crowther 2018, 599–601. University of Massachuses Amherst’s institutional repository.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms References in figure captions and footnotes to specific Consolare), the mid 19th century (e.g., Via Stabiana), iron deposits use the format “ID_[number].” Some and the early 20th century (e.g., Via dell’Abbondanza). deposits contain solid iron remnants. Others are rep- This fact demonstrates that the use of iron and iron resented only by a stain on the paving stones, and we slag was an ancient phenomenon and not an early cannot be certain if these deposits were formed from modern intervention. Moreover, the locations of these iron metal, highly ferric slag, or some combination of ferric deposits are too specific and too frequent to have the two. We therefore refer to deposits generally as been accidental. “iron/iron slag” or “ferric materials” to acknowledge In total, the survey documented 434 occurrences of that ambiguity. Throughout the discussion, we refer ancient iron in the streets at Pompeii. We identified six to the ancient streets by their modern names, and spe- types of deposits that fall across a spectrum including cific locations mentioned in the text are indicated by individual, solid iron objects apparently driven be- number on the map (fig. 2). tween paving stones, solidified slurries of composite materials filling deep ruts or holes, iron stains at the seams between paving stones, and droplets and splat- In July 2014, we conducted a survey over 5.5 km of ters of iron on top of paving stones. Each instance of- Pompeii’s excavated streets, representing 73.5% of the fers evidence for the process of its deposition as well surfaces with visible stone paving.11 In total, we exam- as for the effect Pompeians expected it to have on their ined approximately 17,500 m2 of paved streets (see fig. street surfaces. In the following discussion, we describe 2). The method for this survey was to walk each street, each type in detail and then attempt some quantifica- documenting and describing on a tablet computer any tion of all six types and their distributions. visible traces of iron. Textual and visual descriptions were captured in a FileMaker database and a GPS point Type 1: Solid Without Staining was taken using the Spyglass app.12 Most instances of Within our surveyed streets, there are 132 instances iron were obvious and did not require cleaning be- of solid iron embedded among the paving stones. Al- fore they were documented. For less clear examples, though it can be difficult to distinguish between them, we brushed away grit and debris and, in a few cases, it is clear that workers introduced these pieces of iron poured water onto the street to bring out the iron’s vi- into the street either (1) by pounding a solid iron piece sual qualities of metallic sheen or the rich orange and between the paving stones (fig. 3), or (2) by pouring ocher colors of oxidization. While extensive and rep- molten iron into a cavity, which left a visible, top sur- resentative, our examination was not exhaustive, and face of solid iron (fig. 4). These instances range in size we expect that additional deposits could be found on from very small (ca. 1 cm in the largest dimension in streets with particularly dense iron repairs, such as Via cross section) to very large (ca. 5 cm x 3 cm in cross Stabiana, and on those with reduced visibility, such as section) and vary in shape from roughly circular to Via di Nola, as well as on the streets excluded from this elongated rectangles.13 Unfortunately, it was never pos- survey. It should be noted that there was no correla- sible to examine the depth of these solid iron pieces. In tion between the distribution of iron use and the dates many cases, when the iron itself was oxidized and de- when the streets were first excavated. Iron is found teriorating, it was impossible to determine its original on streets excavated in the late 18th century (e.g., Via shape and size. Conversely, the deteriorated condition and comparatively large size easily distinguished the ancient iron from modern nails that were also found in the streets but were not recorded.14 A few pieces 11 Several streets were inaccessible due to safety concerns or ongoing work by the Grande Progeo di Pompei. of solid iron were clearly driven down between two 12 e Spyglass app was regularly accurate to within a meter, stones, likely to create compression, as documented sucient to place observations in the correct general location and in the order of our survey as we moved down each street. A few, however, deviated dramatically, and we relocated these by 13 Particularly large examples include ID_078, ID_214, their sequence. Because of time limitations, we did not record ID_259, ID_273, ID_374. Particularly small examples include the exact location of 222 points, 91% of which were on Via Sta- ID_067, ID_084, ID_121, ID_185, ID_338. biana. To preserve their general locations, we bracketed these 14 rough the 1990s, prior to the common use of reector- points between known points, which again allowed us to place less theodolites, nails were used in the streets (and elsewhere) as these iron deposits by their sequence. station points for surveying.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms .

. . Plan of Pompeii with streets and points of interest mentioned in the text labeled. by the breakage on each of the stones surrounding the and the solid represent a single event.17 An instructive point of entry.15 This breakage may have occurred at example on Via Stabiana (fig. 6) shows not only how the end of the process of driving the iron into place, the molten iron/iron slag stained the three stones as the hammer likely hit the stones while making the surrounding the rut it was poured into but also how iron flush with their surface. A few solid iron pieces the solid iron reflects the plasticity of its molten state. found lower than the street level might have been The stain’s undulating, flowlike surface adheres and countersunk with another device,16 but often the poor conforms to the particular contours of the lava stones condition of the iron suggests that it has eroded down and the rut cut into them. These examples are clearly to its present elevation. Examples of molten iron can distinct from smears of iron on paving stones, stepping be identified by the irregular shapes of the remaining stones, or curbstones caused by the passage of wheeled solids that conform to the contours of the surround- vehicles with iron tires, even when such smears are ing lava-stone pavers. near solid iron deposits in the street.18 In another example of Type 2 (solid with staining), Type 2: Solid with Staining a dark red-to-purple stain (fig. 7) is preserved only Those examples in which a solid iron mass is found within the small pockmarks on the surface of one accompanied by staining make it abundantly clear that paving stone at the intersection of four paving stones. most iron pieces in the streets, including many Type 1 There are no stains on the surfaces of the other three deposits, were in fact originally poured into place (fig. stones, though staining does appear on a small capping 5). In these 78 examples, an obvious iron stain can be cobble within the intersection of the pavers. A solid observed on the stones adjacent to the solid iron piece remnant of iron/iron slag remains in the junction be- and directly connected with it, indicating that the stain tween the pockmarked stone and the adjacent paver.

15 ID_044, ID_127, ID_307. 17 E.g., ID_046, ID_221, ID_257, ID_280, ID_328, ID_381. 16 E.g., ID_273. It is not certain, however, that this deposit 18 Poehler 2017, 130, and supplemental images, gs. 5.15– was driven in as a solid. 18, at hps://works.bepress.com/eric-poehler/.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms

. . Example of Type 1, solid without staining, Supple- mentary Table, ID_297, Via Stabiana. Note the solid iron at the bottom of a deep rut and degraded iron filling between paving stones.

. . Example of Type 1, solid without staining, Supplemen- tary Table, ID_044, Via Stabiana (ID = iron deposit). The damage to the paving stones around the iron suggests that the iron was pounded into the junction.

. . Example of Type 2, solid with staining, Supplementary Table, ID_328, Via Stabiana. Note the undulating surface of the iron and the adjacent stains.

These observations demonstrate that an iron-rich material was poured on the surface of the pockmarked stone and was then redirected into the interstices among the pavers. One might be inclined to explain the fact that the stain no longer covers the entire area on the top of the paving stone by subsequent rutting from regular use, but the pavement had only recently . . Example of Type 2, solid with staining, Supplementary been repaired.19 It is clear, therefore, that these small Table, ID_381, Via Stabiana. Note the irregular form of the iron and the inconsistency of its oxidization. 19 Poehler and Crowther 2018, online Supplementary Table 1: PE_090 (PE = paving event), hps://works.bepress. com/eric-poehler/102/.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . disprove this hypothesis. Although their colors are not significantly different from ancient stains, these stains are distinct both because they retain the form of the previous objects’ linear and rectangular shapes and because they rest in the middle of a paving stone (fig. 8). Conversely, the irregular shapes of ancient stains supports the notion of their fluid state at the time of deposition. Finally, unlike the modern stains, ancient stainingexcluding droplets or splatters of droplets (discussed below)occur only at the abutments between paving stones and not in the middle of a paver. In the case of iron stains interpreted as pour points, molten ferric materials were apparently poured into gaps between paving stones to repair the spaces between and below them. As the molten material was poured, it covered the adjacent stones and flowed into the gap, but when the material hardened, it did not fill the gap up to the level of the street surface. Such stains occur both in the junctions between two stones and in the intersections of three or more stones. Ancient stains range in color from a light reddish-orange to a dark purplish-gray and differ in intensity and consis- tency from faint enough to be partially transparent to 22 . . Example of Type 2, solid with staining, Supplementary sufficiently thick to retain a metallic sheen. The color Table, ID_157, Via della Fortuna. Note the ferric materials fill- of a stain can differ even within a single pour point.23 ing the junction between paving stones and the stains in the One of the best examples of this comes from a deeply pockmarks on top of one paving stone. rutted section of Via Stabiana where the undulating edges of an orange-to-purple stain can be seen to depressions on the stone left by chiseling to create its flank the junction between two paving stones between flat surface are stained by (rather than filled with) this which the molten material seeped (fig. 9). Although iron material because that material was not allowed to this stain is relatively large at more than 30 cm in set in place. Finally, this solid mass of iron/iron slag length, some stains are exceptional in size, covering five and staining are accompanied by, but not in physi- paving stones and reaching over a meter in length.24 cal contact with, two individual droplets on an adja- Most stains do not cover all the stones at a pour cent stone.20 These droplets (discussed in more detail point. Instead, it is common that only one side of the below) document the events associated with intro- pour point is stained, indicating that a guide was used duction of the iron/iron slag into the street as a liquid. for directing the molten iron/iron slag (fig. 10).25 The flat face of an iron spade would work well not only to Type 3: Staining at Pour Point perform this function but also to push or scrape any It might be argued that iron staining in the street can sufficiently inviscid material on surrounding stones be explained by a solid iron object that, having been into the cavity (as in the example of Type 2, solid with abandoned in the street prior to the eruption (or after staining; see fig. 7). If molten iron/iron slag hardened excavation), oxidized and stained the surrounding on the surface of a street, a spade also could be used to stones. Three modern stains on the paving stones near the Porta Stabia that were created by the placement of 22 21 Examples of reddish-orange stains: ID_027, ID_163, a modern metal stairway, later removed, help test and ID_308; purplish-gray stains: ID_008, ID_231, ID_324; with metallic sheen: ID_001, ID_218, ID_423. 23 E.g., ID_027. 20 ID_158. 24 E.g., ID_322. 21 ID_004–6. 25 E.g., ID_107, ID_191, ID_308.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms

. . Example of Type 3, staining at pour point, Supplemen- tary Table, ID_432, Via Stabiana.

. . Modern example of Type 2, solid with staining, Supple- mentary Table, ID_004, Via Stabiana. The stains were made by the feet of modern steps that have now been removed. break up errant material for collection and reuse. The 70 examples of Type 3 testify to the high liquidity of the material that was poured onto the street and flowed even into small cracks. . . Example of Type 3, staining at pour point, Supplemen- Type 4: Droplet tary Table, ID_008, Via Stabiana. Note that the iron materials The liquidity of the material staining the streets is stain only one stone of the junction. also demonstrated by the 86 instances of individual droplets of iron or iron slag that we found in Pompeii. mer instance, the droplet had either been scraped up We documented droplets in all areas of the streets, at the time of deposition, leaving only a stain, or had including inside deep ruts and on the top of paving been partially worn away by the movement of vehicles stones. This distribution supports the obvious con- over its position.27 In the latter instance, the droplet clusion that droplets were not an intentional part of remained largely in its original form. the deposition of material but are instead accidentally One example of Type 4 on Via del Vesuvio is a drop- produced, and they serve as evidence for that mate- let that had landed on the steeply sloping face of a deep rial’s presence in and movement through the streets. rut. The molten iron material not adhering to the stone Droplets are identified by a red-orange to purple-gray sagged to the bottom of the stain and formed a small stain and/or by a more three-dimensional form accompanied by a metallic sheen (fig. 11).26 In the for-

27 It is conceivable that many droplets were eroded in mod- 26 Droplets with metallic sheen or three-dimensional quality: ern times, which would leave only a stain as evidence of their ID_074, ID_087, ID_316, ID_423. deposition.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms .

. . Example of Type 4, droplet (with staining), Supple- mentary Table, ID_316, Via Stabiana. Note the coronal shape and metallic sheen of the edge of the droplet. lip (fig. 12). Such examples of droplets found inside ruts are especially instructive as they demonstrate that either the duration of time or the volume of traffic was not sufficient to wear away these accidental deposits of iron/iron slag before the eruption in 79 C.E. The position of droplets inside ruts, where the wheels of ve- . . Example of Type 4, droplet (with staining), Supple- hicles would grind over them, would make them highly mentary Table, ID_081, Via del Vesuvio. Note that the droplet susceptible to erosion.28 That they remain, and remain adhered to the vertical sidewall of a rut, and some of the mol- in some cases with little to no evidence of wheel wear, ten material collected at the bottom of the stain in a raised lip. leaves one with the inescapable conclusion that the droplets were landing on the streets in Pompeii in the tion. The 38 examples of splatters appear to result from final weeks or days before the eruption.29 accidental spillage. For example, a collection of droplets could be created by molten material that dripped Type 5: Splatter unintentionally near a pour point. Like water dripping A splatter is defined as either a collection of droplets from a glass that is only slightly tilted, the molten mate- or a large, irregular iron staining not directly connected rial might have run down the outside of its container, to an instance of solid iron/iron slag or a pour point. collected at its base, and then dripped onto the pave- Splatters differ from pour points because they are not ment near the actual pour point. One stain has a par- necessarily located at the seam between paving stones ticularly linear appearance, as though molten iron/iron and because they have more than one point of deposi- slag had dribbled onto the stone as this container was transported above and across it (fig. 13). Alternatively, a large splash of iron might cover a 28 On the interior of ruts: e.g., ID_128, ID_188, ID_211; large number of paving stones. The weight of the ma- on the sidewalls of ruts: ID_071, ID_099, ID_123, ID_162, terial and its container, the difficulty in carrying and ID_174, ID_319, ID_367. One droplet was even found on top of a stepping-stone: ID_229. manipulating them, and the heat of both would have 29 Unsurprisingly, Pompeii produces many examples of evi- been hard enough to manage even if they did not have dence for activities ongoing at the moment of the eruption. For to be transported over deeply rutted streets. Some- example, cuing beam pockets (Quadriporticus, VIII 7, 16; times, it seems, the ironworkers simply spilled their Poehler and Ellis 2013, 11), painting in a house (Casa dei piori load. Large splatters of iron staining are rare, as any ac- al lavoro, IX 12, 9; Varone 1998), bread le in the oven (Casa cident should be, but it is perhaps surprising that trips del Fornaio [Modestus], VII 1, 36; Benton 2014, 80, 266–68), and the paving of streets (Vicolo del Conciapelle; Poehler and and slips were not more common. Indeed, the largest Crowther 2018, 596–97). concentration of splatters was spilled over at least five

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms section of Via Stabiana not been disturbed, it might have been possible to deduce more precisely how that event transpired.

Type 6: Repair with Composite Materials In 29 instances, the quantity of the iron and iron slag in the street is far greater than in any individual instance among Types 1–5. It is clear in these examples that the intention was to repair the deeply rutted and failing stone surfaces using molten materials. Undoubt- edly, the most dilapidated sections of street in Pompeii are on Via della Fortuna west of its great intersection with Via del Vesuvio, Via di Nola, and Via Stabiana . . Example of Type 5, splatter (with staining), Supplemen- where most examples of such large-scale repairs are tary Table, ID_429, Via Stabiana. Note the linear arrangement found (see figs. 2[2]; 14). One particularly long rut of staining and droplets across the paving stone. along the north curb was filled by several meters of iron repair materials, of which significant portions survive for examination (fig. 15). Several features are stones in a well-rutted (and subsequently disturbed) immediately apparent. First, iron and iron slag were section of Via Stabiana. These stains, some still retain- not the only materials used in the repair. In every ex- ing a clear metallic sheen, cover comparatively large ample, additional materials were used, including stones portions of the paving stones and appear to represent of varying sizes, ground-up pieces of terracotta and a particularly egregious accident and not a large pour ceramics, and in some examples, small pieces of white point.30 limestone. Second, the iron substances were poured di- Unfortunately, this section of Via Stabiana was rectly into the rut, covering it throughout its length but significantly disrupted by the aqueduct built by Do- not deeply in all places. In fact, in one rut, as it passes menico Fontana in 1592, which cut directly across over the top of a single stone, the preserved height of the street (see fig. 2[1]).31 Additionally, interventions the material is little more than a centimeter or two in for the infrastructure of the modern tourist site have ruts that are 10–20 cm deep. This indicates that the disturbed this part of Via Stabiana with the removal objective was to prevent further damage, not to fill of some ancient paving stones, the addition of some the rut completely and restore a flat surface. On the modern stones, and the repositioning of some ancient other hand, it seems that these materials were meant to pavers without regard to their original organization.32 penetrate more deeply into the gaps below the surface Nonetheless, even in their disturbed state, these re- and between paving stones that were nearly or entirely maining ancient pavers record an event, or perhaps cut through by ruts. Specifically, with these repairs of several, in which a large amount of molten, ferric mate- composite materials in place, something of the original rial was accidentally deposited onto the street. Had this compression between paving stones was reproduced, and at the same time, a kind of bridge was created for vehicles to cross over the now-filled gaps. 30 ID_426, ID_427, ID_429, ID_431. 31 Images from the late 19th and early 20th centuries show Even if we allow for a layer of repair only 2 cm deep the external curve of a vaulted channel that Fontana had cut within a much deeper rut and estimate that 50% of its through Via Stabiana prior to its excavation. See Poehler and volume consisted of nonferric materials, the volume of Crowther, 2018, online Supplementary Figure 1: hps:// ferric material is considerable. Our calculations indi- works.bepress.com/eric-poehler/99/. e same channel is still cate that more than 70 liters of molten iron/iron slag in place across Via di Nocera in the eastern portion of the city. See Via di Nocera street views: hp://pompeiiinpictures.com/ was poured onto this section of Via della Fortuna to 33 pompeiiinpictures/Streets/Via_di_Nocera p2.htm. repair it. How deeply these repairs might have pen- 32 In this area, some paving stones have a rut on one edge, but the adjacent stone does not have the other half of that rut. In other examples, ruts now run (impossibly) perpendicular to the 33 is calculation is based on an estimate of the area of the street. street rued between the stepping stones in the middle of this

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms .

. . Deeply rutted section of Via della Fortuna containing remnants of many repairs. etrated below the surface when filling a pothole or a one of Pompeii’s first lava-stone surfaces.35 Because the large gap between the stones rather than a rut in the focus of the report was on the efforts to improve the street’s surface can be explored thanks to an excava- surface drainage of the modern site, a detail crucial to tion beneath Via Stabiana just south of its intersection our discussion was omitted: more than 10 rectangular with Via del Tempio d’Iside and Vicolo del Menandro storage trays (or ca. 33 liters) of ferric, composite repair (see fig. 2[3]). In 2011, a short report was published materials were recovered and preserved from an exca- on work by the superintendency to capture and redi- vation less than 4 m2 in area.36At the time of writing, rect rainwater from Via Stabiana into the Domenico these materials had not yet been studied by a specialist. Fontana aqueduct, thereby transforming the early Visual examination of the contents shows that a signifi- modern water channel into the present site’s sewer.34 cant amount of iron/iron slag was poured into a cavity Interestingly, at approximately 80 cm below the cur- that was already partially filled with a number of other rent surface, the excavators found an earlier lava-stone materials, including relatively large fragments of lava surface, which, as we have argued elsewhere, was likely stone, terracotta and ceramics, and several instances of

segment of paving (segment identied in Poehler and Crowther 2018 as PE_129). Measurements between the stepping-stones 35 Poehler and Crowther 2018, 588–89. and the curbs were multiplied by a length of 10 m to nd the to- 36 e authors are grateful to Do. Catello Imperatore for tal rued area, which was then multiplied by a percentage of the alerting us to the existence of these materials from his excava- area that was deeply rued. For example, the area between the tion. Our calculation of volume is as follows: 10 trays (55 cm center and southern stepping-stones was 49 cm wide and, over x 33 cm x 11 cm), lled two-thirds full contain 133.8 liters. We the course of its length, was deeply rued over 50% of its area. estimate that the material is 75% nonferric ll, which leaves 33.4 34 Rispoli and Paone 2011. liters of ferric materials.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms

. . Example of Type 6, repair with composite materials, Supplementary Table, ID_168, ID_169, Via della Fortuna. Note that the deposit contains ceramics, including terracotta, as well as pieces of marble or white limestone, and it fills only the bottom of the rut. what appears to be copper based on the familiar green tint of its oxidized form.37 This excavation reveals the depth to which some of the iron/iron slag fills penetrated below the street surface as well as the very large quantities of material deposited. Indeed, only a few meters south of the superintendency’s trench, another repair can be seen filling a large gap between the paving stones, and we can conjecture that the material of the repair continued for some distance below (fig. 16). It is probable that the depth of the repairs on Via Stabiana is unusual, and we should not think that such great volumes of molten material disappeared into the seams between paving stones wherever there is a stain nor that each . . Example of Type 6, repair with composite materials, solid iron piece is the tip of a much larger iceberg.38 Supplementary Table, ID_422, Via Stabiana. Note that the Nonetheless, the volume of material at only these two deposit (ca. 40 cm x 20 cm) includes large stones and that it locations (i.e., Via della Fortuna and Via Stabiana) fills a large gap between paving stones. in addition to the wide distribution of the supporting evidence in Types 1–5demonstrates the commit- ment Pompeians had made to the use of iron for street repairs and the scope of the problems they had hoped to solve by it. 37 Slag heaps from Populonia show layering of iron (Roman) above copper (Etruscan) smelting activities (Benvenuti et al. Distribution and Quantification 2000). Mixing of these slags might account for the presence of Having described the deposits of iron material in copper in these deposits. the streets of Pompeii, we now turn to more quanti- 38 at these materials were poured into Via Stabiana rather tative and spatial analyses to reinforce the four main than used as ll in the construction of the street is supported by points introduced above, that (1) the iron materials in staining elsewhere on this street and by the absence of slags being used as a bedding layer anywhere else in the city, including the street are ancient; (2) most were applied in liquid in the repaving of Vicolo del Conciapelle, which was ongoing in from; (3) their purpose was to repair the streets; and 79 C.E. Additionally, it is tempting to wonder if these deep lls (4) many (if not all) repairs were made not long before of iron materials below Via Stabiana relate to the earlier street the eruption of Mount Vesuvius in 79 C.E. Moreover, surface having acted as a barrier to water soaking into the soils, we further elaborate on the notion of street repair, as which consequently eroded the lls supporting the nal surface.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . these analyses reveal two different mechanical proper- iron tangs were wedged between the pavers (some- ties understood and deployed by the Pompeians. times damaging the stones in the process) to increase Support for all four points comes from examining the pressure across the surface of the street. Second, the specific locations of deposition, which show where and most frequently, Pompeians poured liquid iron iron materials were used and where they were not, as materials to fill holes and gaps, especially into those well as how different types of iron materials were em- spaces that could otherwise be reached only by remov- ployed. At any particular location, we found iron de- ing the paving stones, and so bound the streets below posits on a single stone, in the junction between two by adhesion as well as compression. In perhaps the stones, or in the intersection of three or more stones. most dilapidated street section, the eastern portions of When combined with the type of iron material pres- Via della Fortuna (see fig. 2[2]), ruts had almost com- ent, detailed information about the location of the pletely cut through the paving stones, and it is here that deposits offers important insights. First, more than just over half of all repairs using composite materials two-thirds of the observations record the presence were found.41 But even those ruts that contain only a of liquid iron through staining evidence, even though small amount of iron reveal a strong relationship be- solid iron without staining (Type 1) is the single larg- tween ruts and iron materials: 21% of all iron and iron est category at 31% of the total. Solid iron pieces were staining were found specifically within a rut, but more most commonly found in the intersections among than 84% of all iron deposits were found in street sec- stones, while staining was most often identified on a tions having deep ruts (as defined by Tsujimura; table single stone. 1). Moreover, in streets with little to no rutting,42 66% The specific locations of iron staining reveal how of the iron deposits identified were droplets or splat- carefully (despite the minor loss of material from drop- ters rather than repairs, probably the result of spills lets and splatters) the Pompeians seem to have poured during the transit of molten iron material to locations this material. When pouring iron/iron slag at an inter- with deep ruts.43 A related correlation is the strong re- section of three or more stones, they stained only one lationship between the age of the pavement and the stone in 51% of the deposits, two stones in 37%, and likelihood of its having received iron repairs: 65.1% three or more stones in only 12%. The percentages of all iron deposits are found within sections of paving are even stronger at pour points identified at the junc- that were more than 79 years old at the time of the 79 tions of two stones onlyseven out of 10 times they C.E. eruption (table 2). stained only one of the two stones. As argued above, These statistics demonstrate that the Pompeians these figures indicate that some kind of guide, perhaps targeted deeply rutted streets, regardless of age, for this a spade, was used when pouring the material. Such pat- novel repair process. Additionally, the fact that some terning in the placement supports our belief in both iron repairs were applied to streets paved or repaved in the antiquity and liquidity of the material. That is, if the last decades before the eruption shows how late in such staining were to be imagined as iron objects left in the life of the city this process was introduced.44 For ex- the street to corrode away over decades or centuries,39 ample, the short stretch of pavement on Via della For- it stretches credulity to explain how such lost items tuna (18.5 m) between two particularly deeply rutted would have92.8% of the timefallen or been placed sections, which represents a repair made in the period not only at the seams between paving stones but also on alignment with those seams.40 The consistent po- sitioning of staining is further supported by the 132 41 ID_141, ID_142, ID_147, ID_148, ID_168–71, ID_173, ID_175–79, ID_182. solid iron objects, all of which are found in the seams 42 between paving stones. at is, in Tsujimura’s (1991) shallow, faint, and no rut categories. These data also demonstrate that two different con- 43 e likelihood that deposits were unintentional spills, cepts of force were applied to repair Pompeii’s streets. rather than repairs, decreases as rut depth increases: droplets First, where greater compression was needed, a few and spaers account for 71% of deposits in streets with no ruts, 61% in those with faint ruts, 50% in those with shallow ruts, and only 23% in those with deep ruts. 44 It is conceivable, of course, that this use of iron was applied 39 For example, modern stains (supra n. 21). in early periods and continued until the eruption. ere is, how- 40 Of the 70 examples of Type 3 (staining at pour point), only ever, no evidence to support such an argument, and that sup- ve appear on a single stone. position is made less likely due to the consistent use of stone for street repairs in previous eras.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . Iron Deposits by Depth of Ruts (as defined in Tsujimura 1991). No. of Iron Deposits by Rut Depth Iron Deposit Type Deep Shallow Faint No Rut Type 1: Solid without staining 123 3 3 3 Type 2: Solid with staining 75 2 – 1 Type 3: Staining at pour point 58 6 6 – Type 4: Droplet 52 13 13 8 Type 5: Splatter 34 2 1 1 Type 6: Repair with composite materials 25 4 – – Total 367 30 23 14 % of all iron deposits 84.6 6.9 5.3 3.2 following the earthquake of 62/3 C.E., received at least . Iron Deposits by Age of Pavement nine iron deposits that must date after that event.45 But (using phases of paving activity in Pompeii as defined it is the small stains and droplets within rutsdeposits in Poehler and Crowther 2018). that could easily have been eroded away by the passage No. of Iron % of All Iron of vehiclesthat indicate that this remarkable experi- Phase of Paving Activity Deposits Deposits ment was likely still occurring in the last weeks if not days before the eruption in 79 C.E. Phase 1 (100–20 B.C.E.) 16 3.9 ? Phase 2 (20–1 B.C.E.) 253 61.3 The evidence for repairs using molten iron mate- Phase 3 (1–20 C.E.) 47 11.4 rial raises technical questions. It has been, and indeed Phase 4 (21–40 C.E.) 56 13.6 remains, a common assertion that Romans lacked the technical ability to reach the temperatures required to Phase 5 (41–62/3 C.E.) 32 7.7 46 liquefy iron, especially in its pure forms. This ques- Phase 6 (62/3–79 C.E.) 9 2.2 tion was a subject of intense debate in the first half of the 20th century, and skeptics pointed to the general absence of both the blast furnace and cast-iron objects These empirical and theoretical concerns have come from the archaeological record.47 Equally important to be better understood in recent decades as new ex- were the assumptions underlying these arguments, in- cavations, more focused metallurgical study, and excluding a unilineal model of technical evolution and a perimental archaeology have been brought to bear on rejection of the idea that Romans could produce cast questions regarding molten iron in antiquity. iron because they rarely used the resulting metal.48 The simplest challenge to dispel is the claim that there is an absence of cast iron in the archaeological re- 49 45 cord. More than a century ago, indisputable examples Poehler and Crowther 2018, online Supplementary Table of cast iron were recognized and published from Hen- 1: PE_090. e iron interventions on this paving event, how- 50 ever, are found near to the repairs of the particularly damaged gistbury Head and Warrington in England. These sections: seven of nine iron deposits (ID_151, ID_153–55, early objects and excavated furnaces required the ad- ID_157–59) are located within a few meters of the deeply rutted sections, which had dozens of iron deposits in each. 46 Supra n. 9. 47 linear “evolution of the metallurgical furnace.” Craddock (1995, 259) notes that ironworking furnaces of 49 For the most recent review of the evidence, see Pleiner all types are rare in the archaeological record. 48 2000, 247–49. Craddock 2003, 231. See Forbes (1966, 78, g. 14) for the 50 May 1904, 26; Gowland 1915, 76–7.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . mission that melting iron was not impossible using Experimental archaeology has added additional Roman technology, and May went so far as to argue relevance to these pieces as well as those found on that cast iron was being intentionally produced and slag heaps by demonstrating that the Roman shaft refined.51 Unfortunately, subsequent scandals of mis- (bloomery) furnace could produce cast iron with only identification52 and outright forgeries53 led the schol- slightly more human effort. Reconstructed furnaces arly community away from the question. Throughout have shown that Romans did regularly and necessarily the middle of the 20th century, it was the communis achieve temperatures above 1400°C, sufficient to liq- opinio that “cast-iron must be considered unknown to uefy some iron ores, and could surpass 1600°C, which Antiquity,”54 and even direct references to iron cast- is beyond the melting point of pure iron.61 More recent ing by Aristotle, Pausanius, and others were explained experiments indicate that “cast iron should be regarded away.55 Scholars tended to dismiss the few unimpeach- as an inevitable by-product of bloomery smelting,”62 able artifacts that were discovered as the by-products which “was easily made if wanted,”63 and not the result of an accidental overheating, an idea bolstered by the of a mishap. It is, moreover, not necessary to conclude repeated claim that most cast iron was found in slag that cast iron found in slag heaps was regarded as use- heaps because it was deemed useless.56 Yet, this narra- less by Roman ironworkers. According to Crew et al., tive sets aside the fact that some of the earliest items remains of cast iron instead usually would have been known were not discarded lumps of misfired bloom.57 recycled in subsequent smelts, and pieces recovered Intentionally formed objects include a cauldron rim in slag heaps only represent the remains that smelters fragment from Brading Roman Villa on the Isle of failed to reuse.64 This observation is significant as it Wight, a small block of cast iron from Wilderspool suggests that our corpus of cast iron finds both does (Warrington), and a small iron bar from the Roman not reflect the amount of this material produced by the industrial area at Tiddington.58 If this last piece was a Romans and masks their contemporary understanding billet, it could support the Roman origin of larger cast- of its significance for their activities. By the end of the iron ingots from Dordogne, France, the date of which 20th century, Tylecote could state that examples of are in question.59 Even those scholars who accepted Roman cast iron are “innumerable” (though still de- this evidence for intentional objects, such as Aitchison, nying their significance), and Pleiner could list eight continued to argue that cast iron was unintentionally objects of intentionally produced cast iron.65 produced and resulted from the “freakish behavior” of Today, the claim that Romans could not melt iron some “run-away” furnaces.60 cannot be sustained.66 Moreover, the claim miscon- strues Roman technical preferences as technical incompetence. Cast iron was too brittle for most pur- 51 May 1904, 21–9. poses, so the idea in smelting was to cause the iron to 52 On the so-called Hallsta ring, see Coghlan 1977, 49. bloom, not to liquefy. It was not the case that Romans 53 For example, Dawson’s Beauport Park Statue. Charles could not create cast iron; it was instead their common Dawson was a master forger and perpetrator of the great Pilt- intention not to do so. For this reason, current discus- down Man hoax. For the story of the forgery, see Russell 2003, sions of Roman metallurgy more often focus on why 34–8; for metallurgical analysis see Craddock 2009, 477–79. 67 Coghlan (1977, 49) lists other statuees that have been found they preferred the bloomery process, their experi- and discounted because of poor provenance. 54 Forbes 1950, 407. Aitchison (1960, 205) and Schubert and Ingen Housz (1957, 57), among others, repeat this formulation. 55 Pleiner (2000, 139–40) discusses the ancient sources ex- 61 See Morton and Wingrove 1970, 6; olander and cept Pausanius and Zosimos, for whom see Forbes (1950, 408) Blomgren 1985, 417; Pleiner 2000, 133–37 (citing olander and Craddock (1995, 279; 2003, 243–44), respectively. 1987); Sim 2011, 85–6. 56 Collingwood and Myres 1936, 233; Forbes 1950, 407–8; 62 Crew et al. 2011, 258. Although these experiments repli- Craddock 2003, 234; Rihll 2013, 57–8. cated a 14th-century furnace, its design is unchanged from Ro- 57 Coghlan (1977, 47–9) admits these as examples of inten- man bloomeries. tional Roman cast iron while maintaining their unimportance. 63 Rehder 2000, 127. 58 Fieldhouse et al. 1931; Cleere 1958, 72–4; Aitchison 1960, 64 Crew et al. 2011, 258. 205, table 19. 65 Tylecote 1986, 168 (followed by Craddock 2003, 249); 59 Wertime 1961, 46. Another pig-iron lump is catalogued at Pleiner 2000, 248–49. Bonn: Meier-Arend 1984, 368 n. 74. 66 Note the absence of this claim in Lang 2017. 60 Aitchison 1960, 205. See also Coghlan 1956, 73. 67 Rehder 2000, 141.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms ments within and beyond it,68 and what such empirical a crucible furnace, perhaps one not unlike that shown knowledge might have taught them in the absence of in the frescoes from the Casa dei Vettii (fig. 17)73 at (modern) theoretical constructs.69 Pompeii or the Catacombs of Domatilla74 at Rome. If most Roman smiths lacked a technical ability, it It was not the primary action of reducing raw ore into was how to decarburize the cast iron they regularly an iron bloom. This secondary process did not involve produced to make it less brittle and so appropriate for concern about overheating since melting and even use as tools. Once again, however, there is a growing superheating the material was the desired result. Sec- body of evidence that some Roman blacksmiths were ondly, the Pompeian workers might have used many learning how to refine high-carbon and cast irons for varieties of iron and iron-rich materials with a signifi- use.70 There are today a few indisputable examples of cantly lower melting point than pure iron. Some iron evidence and an evolving sense that these practices ores, such as bog ores (limonite) and lake iron, are were more common than previously understood.71 It rich in phosphorus and consequently have a melting is clear that there was a range of Roman ironworking point closer to 1200°C. Similarly, cast iron and iron- practices and competencies as well as many socio- smelting slags have a melting point between 1100°C economic incentives and disincentives to employ and 1200°C, which makes both good candidates to be them.72 It is worth pointing out that we see such varia- identified as the materials in the streets at Pompeii.75 tion in the application of many Roman technologies, Since slag is a waste product and cast iron is a less de- with cutting-edge and lagging examples often exist- sirable (if recyclable) by-product of smelting than pure ing side-by-side. For example, Pompeii’s architectural iron, these lesser materials would also have the benefit environment had long benefited from the pozzolanic of cost effectiveness in addition to their lower melting bonding agents locally available for use in mortars, yet points, a point we will return to below. at the time of the eruption, few buildings had taken full advantage of the boom in cement construction that was revolutionizing Roman architecture at Rome ? and beyond. How the Romans introduced liquefied iron mate- While it is now beyond doubt that the Romans rial into the streets at Pompeii remains a mystery. The could and did melt iron and were perhaps learning process must have included obtaining the necessary to use the resulting products, it is important to re- raw materials (e.g., iron and iron slag, charcoal), as- member that the process of pouring a molten ferric sembling the facilities and tools (furnaces, crucibles, material into the streets at Pompeii did not necessar- tongs, hammers), and transporting the liquefied ily require the same level of effort or expertise. In the materials (vehicles, manpower, insulating and pour- first instance, the process at Pompeii was almost cer- ing devices). Fortunately, archaeological research on tainly a secondary event of melting some material in Pompeii’s industrial landscape and on the primary metal-producing regions of the Roman world allows an attempt at reconstructing the process of making 68 Fluzin 2000; Craddock 2003, 249–51; 2008, 109; Cech iron repairs at Pompeii. To be clear, our description is and Rehren 2014. See especially Lang’s (2017, 863) discussion speculation, but, like any good argument built on cir- of Roman steel. cumstantial evidence, it puts the plausible, if not the 69 Crew 2013. Stewart (2002, 114) remarks that “the ability actual, in place of the absent. of pre-historic and Roman smelters to feel the way empirical- Unfortunately for our reconstruction, metalwork- ly towards such results, without knowing the reasons, is to be ing at Pompeii is a surprisingly understudied subject, respected.” 70 Craddock 2008, 107–8. especially when it comes to iron. In 1988, Gralfs pub- 71 See Crew et al. (2011, 258) for additional bibliography. lished her thesis on the 11 identifiable sites of metal 72 Earlier scholars deserve credit for expecting such variabil- ity from limited evidence. As early as 1950, Forbes (1950, 407) pointed out that “even in the sixteenth century primitive and more sophisticated methods of metalworking were used side 73 See Monteix (2016, 204–5) for the most recent technical by side and dierent forms of smelting furnaces were adapt- description. ed to the peculiar characteristics of local ores. No wonder that 74 Kelleher 1982, 225, no. 144. our picture of Roman and pre-Roman iron smelting is far from 75 On the melting points of iron and slag, see Cleere 1971, clear.” 205; Rehder 2000, 103–12; Stewart 2002, 91, g. 4.22.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . front room,79 and “a formless mass of iron”80 that Gralfs believes could have been slag,81 the scale of the process to repair Pompeii’s streets with iron is far beyond the capacity of this locale. Because Gralfs could identify only two other primary ironworking facilities within Pompeii, both of which were still smaller in size,82 the production site for the street repairs must have been located elsewhere, probably beyond the city walls. An extramural location fits the pattern for other heavy industries at Pompeii. For example, the pottery industry at Pompeii in 79 C.E. was limited to two workshops at the urban periphery, one inside the Porta di Nocera (I 20, 4) and the other outside the Porta Ercolano (see fig. 2[5, 6]).83 While these facilities appear to have been capable of serving Pompeii’s daily ceramic needs, none of the kilns known at Pompeii were capable of creating larger vessels such as ampho- . . Wall painting depicting a crucible furnace, Casa dei ras and dolia. Still, fabric analysis of stamped ampho- Vettii, Pompeii, Room q, south wall (© DAI Rome, neg. D- ras and tiles indicate that, at least in the last quarter of DAI-ROM 31.2736). the first century B.C.E., the workshop of L. Eumachius was producing these ceramics, though its location is 84 production within Pompeii, proving that the city’s unknown. If these larger materials were produced capacity was far greater than previously understood. locally, they must have been fired at an external pro- 85 Still, metalworking was not a robust industry and op- duction facility. erated mainly as small businesses.76 On the other hand, A similar phenomenon can be identified in Pom- some of the tools demonstrate that Pompeian smiths peii’s fish-salting industry. Although Pompeii was were apprised of cutting-edge techniques and that local famous for its garum production, nearly every street- consumption could support some larger-scale bronze front facility with a fish-processing vatand there producers.77 Perhaps the most illustrative of these must have been scores of them spread across the city small-scale smithies is the so-called fondo Barbatelli, had been transformed by the beginning of the first cen- 86 located just outside the Porta del Vesuvio (see fig. tury C.E. into a food and drink shop. Yet the garum 2[4]), where smiths produced a wide range of objects industry did not disappear along with its small-scale, in bronze, iron, and lead including statues, serving ves- intraurban production sites. In fact, one leading indus- sels, and furniture parts.78 Although it was never fully trialist, Umbricius Scaurus, continued to profit from excavated and was subsequently reburied, the organi- fish sauce, exporting it around the Mediterranean until zation of space, the metalwork implements, and the objects produced at the “fondo Barbatelli” represent the wide technical abilities of the Pompeian metal 79 Sogliano 1900, 599. industry and the limited scale at which it operated. 80 Sogliano (1899, 444): “un amasso di ferro informe.” 81 Gralfs 1988, 18. Therefore, this workshop cannot be the site of produc- 82 tion for Pompeii’s iron street repairs. Even though the I 6, 1, and VI 3, 12–13. 83 A comprehensive overview of the evidence for the nal de- workshop had ironworking tools, a crucible forge in its cades at Pompeii can be found in Peña and McCallum 2009a, 64–76. A third facility is possible, though unlikely, at II 3, 8. Plac- ing kilns in marginal locations seems to have been a feature of the early city as well. See Peña and McCallum 2009a, 58; Dicus 76 Gralfs 1988, 107. For a brief review of ironworks, see Plein- 2014; Ellis et al. 2015, 2–5. er 2006, 139. 84 Peña and McCallum 2009a, 77; 2009b, 176–79. 77 Gralfs (1988, 71–6, 102–3) includes two workshops in her 85 It is possible that a large production facility remains buried discussion of the Casa delle Forme di Creta: VII 4, 59–60, and in the unexcavated areas of Pompeii. VII 4, 61–2. 86 Ellis (2011) discusses ve vats in his excavations and re- 78 Gralfs 1988, 12–48. ports 11 more vats in Pompeii, all out of use in 79 C.E.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms Pompeii’s final days.87 Once again, we have a market for This high ratio was required within the shaft furnace the products and evidence of the products themselves in order to separate the iron from the rock. Melting but not the site of their production. iron in a crucible furnace would have been more effi- Therefore, if Pompeii had a large iron-production cient. If melting the iron required only half the amount facility, or if a new site was created specifically for the of charcoal, the 6.65 kg billet described above would repair of streets, it almost certainly had to be located require 20–33 kg of charcoal, or 6,000–9,990 kg for outside of the city. Whether there was one large facil- all the observed deposits. In terms of space, each melt ity or several smaller ones, any such production area would require a cube of charcoal approximately 50 cm would have needed areas dedicated to the storage of on a side, or at least 29.7–49.0 m3 (a cube ca. 3.0–3.5 m (1) the iron materials; (2) charcoal for fuel, which on a side) for all the observed deposits.93 None of the might have itself been produced there rather than pur- known locations of metalwork seems to have sufficient chased requiring still more space; (3) the waste mate- facilities (furnaces), space (dedicated storage areas), rials produced; and (4) the vehicles and animals that or forms of access (ramps, stables, and corridors) for would transport all these materials. Although we do these processes, even if all these materials were not not yet know the full volume of materials poured into present at once. the streets, some notion of the scope of the project can Still, the furnaces could not have been located very be gained if we assume some minimum figures for the far outside the city because the most difficult aspects iron deposits. For example, if each of the 309 locations of this operation would have been transporting a sub- where iron materials (including solid iron but exclud- stantial volume of molten iron material into Pompeii, ing droplets and splatters) were identified represents maintaining its heat (and thus its liquidity), and then only one deposition of an average-sized, melted iron pouring it out at specific locations across the city.94 billet,88 such as those found at Newstead (lgth. 30 cm Even if the foundry were located only 150 m outside x diam. ca. 6 cm, wt. 6.65 kg)89 or those from the ship- the Porta del Vesuvio near the “fondo Barbatelli,”95 the wrecks at Saintes-Maries-de-la-Mer (lgth. 29.5 cm x distance from the Porta del Vesuvio to some of the wdth. 10.5 cm x ht. 5.2 cm, wt. 6.79 kg),90 then no less nearest concentrations of deposition in the streets was than 2,000 kg of iron was needed. In some cases, the 250 m farther, with the most distant being over a kilo- amount of iron that is currently visible suggests that meter away. Conveying these materials at a moderate less than an entire billet was required, but many of the walking pace (3.2 km/h) would require approximately iron stains indicate that at least this amount was used. 10 minutes or more to reach most places in the city. Given what we know from excavations below Via Sta- Since the difference between the ambient temperature biana discussed above where perhaps 30 times the and that of the molten metal will always be more than volume of a single billet was deposited at one location, 1100°C, any molten material exposed to the air rapidly this estimate for the total amount of ferric materials begins to solidify outside of the furnace.96 To maintain used at Pompeii might be quite conservative indeed.91 its liquidity, which would be close to that of water, the Similarly, although melting cast iron or iron slag re- molten materials would likely have been both super- quires a lower temperature for a shorter duration than heated (raised to temperatures above the melting point does smelting iron from raw ore, the estimated fuel requirements and heating times would still have been significant. For smelting iron at Populonia, Saredo 93 ese calculations are based on a specic gravity of 208 kg/m3 for charcoal. Parodi estimates that between six and 10 kg of charcoal 94 were needed to reduce a single kilogram of iron (i.e., We considered the possibility that the Pompeians had used 3 92 a mobile furnace, but we have not pursued the idea because of 127 cm , which is a cube merely ca. 5 cm on a side). a lack of evidence and because the volume of such a small-scale furnace might have been insucient to produce the documented repairs. 87 Curtis 1988; Ellis 2011. 95 e distance to the poery facility outside Porta Ercolano 88 On an “average-sized” billet, see Tylecote 1986, 167–68; is 167 m. 1987, 254–55; Pleiner 2006, 40–3. 96 Solidication rates of metals, as dened by Chvorinov’s 89 Curle 1911, pl. 65. Rule, are a function of temperatures, properties of the metal and 90 Pagès et al. 2008, 265–66, gs. 2, 3; Type 6C, SM6_10. mold materials, and mold shape. e average rate for steel to be- 91 Supra n. 35. come completely solid is expressed as four minutes per square 92 Saredo Parodi 2013, 39, 42. centimeter of mold surface area.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . to compensate for heat loss in transit) and insulated (to prevent heat loss in transit). At a minimum, a lid for the crucible would be required to prevent rapid solidification of the exposed top surface of the materials.97 Maintaining fluidity, however, is not the only reason to insulate the crucible. To protect the workers during transport and pouring, it seems very likely the crucible was placed within another container such as a modified transport amphora, perhaps partially filled with lapilli or sand. In this reconstruction, the total weight of the melt of a single billet of 6.65 kg (850 cm3) in a ceramic crucible of 2 kg placed in an insulating, sand-filled container weighing 10 kg would be 18.7 kg, which is manageable to transport and manipulate in pouring. Yet, several of the repairs of composite materials are so large that . . Terracotta plaque at VII 4, 16, in Pompeii depicting por- they might require dozens of trips if each pouring was ters carrying a large amphora on a beam. Note that the porters less than a single liter. Multiplying the volume of the use staffs, apparently for balance. molten material even by five would increase the load to over 45 kg, a weight that would begin to test the strength of a single man, and a tenfold increase would it is clear that some type of guide was used to direct tax even two men.98 For comparison, consider that a the iron materials into the cavities below and between Dressel 20 amphora filled with oil weighs about 105 kg the paving stones. At the same time, the large number and carrying one is depicted by the Romans as a two- of droplets and splatters demonstrate that pouring man job (fig. 18).99 This comparison, however, leaves from the crucible was a challenge that made the small out of consideration the very high temperature of the losses of material common and larger accidents not iron materials, which, even when insulated, would re- uncommon. quire porters to keep their distance. It therefore seems There are no firm answers about how the iron re- best to assume that, regardless of weight, insulated pairs were made, as there is no direct evidence be- crucibles were carried on poles between a pair of por- yond the repairs themselves. Nonetheless, we have ters in a manner similar to the representations of other attempted to circumscribe the possibilities in the con- objects being carried. text of what is known. The economic forces acting on Finally, using the evidence available, we can also other heavy industries suggest that our iron facilities begin to reconstruct how the pouring was accom- were extramural. The high weight, temperature, and plished. It is first important to note that, when molten, cooling rate of the iron materials further suggest that metals are highly inviscid, flowing almost as easily as individual deliveries were of limited volume and that water. Molten slags are also quite fluid, even at lower the process of pouring was difficult. temperatures; slags are as much as five times more 100 fluid than motor oil at 1200°C. These facts fit the ? evidence in the street that the ferric materials flowed quickly into desired locations and at times escaped to Despite hundreds of observations of iron in the form accidental depositions. As mentioned previously, streets, the technical capabilities of the Romans, and a plausible reconstruction, the uniqueness of this phenomenon drives the skeptical mind to search for a 97 On crucibles, see Forbes 1966, 75–6; Tylecote 1986, 95– simpler solution. Why did the Pompeians use iron ma- 101; 1987, 183–92; Bayley and Rehren 2007. terials to repair their streets? To answer this, we must 98 Rihll (2013, 58) suggests that 50 kg was the limit for metal- first better understand the circumstances in which they workers to li and manipulate. She estimates 15 kg for the cru- employed this novel procedure. There must have been cible and pouring apparatus. 99 Peña 2007, 305. some reason for rejecting the ordinary ways of main- 100 Rehder 2000, 111. taining their streets in favor of a method that was per-

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms haps completely unprecedented. That impetus is best more intense than in the previous 17 years and more found in the history of street repairs at Pompeii and in than double the average of the last century.104 Equally the cost in time, money, and opportunity involved in important, however, are the streets that were not repaving and repaving lava-stone streets. paved even though some sections were in desperate Beginning in the late second or early first century need of repair. Via Stabiana was nearly cut through by B.C.E. and continuing until the city’s destruction, ruts, and the easternmost section of Via della Fortuna Pompeians transformed their beaten-ash streets (bat- was more rut than road. Why these streets languished tuti) into a network of lava-stone pavements that over in the plans for street repair is best explained, ironi- time became deeply rutted. The process of paving cally, by their essential nature in the street network. Pompeii was gradual and, though characterized by In fact, the intersection of Via Stabiana and Via degli periods of intense activity, was never fully completed. Augustali/Via Mediana, near the center of the city, is More than half of the lava-stone pavements were laid the most heavily trafficked location in the entire city. for the first time in the Augustan period, probably to- A recent network analysis of pedestrian movement at gether with the water system, as part of a grand infra- Pompeii shows that more than 91% of all the possible structural project.101 Short extensions of lava-stone routes across the city use this intersection. The four- pavements off the main thoroughfares and onto side way intersection of Via del Vesuvio, Via della Fortuna, streets indicate that Pompeians expected to expand Via Stabiana, and Via di Nola, ranked in the top 15% these pavements. Instead, the need to repair the main of streets in volume of traffic, is not far behind.105 It streets diverted attention and resources during the is in the Via del Vesuvio, Via della Fortuna, and Via middle of the first century C.E.; less than a quarter of Stabiana that the most intense use of molten iron/ all the pavements laid between 20 C.E. and 62 C.E. iron slag was documented, amounting to 81.1% of all expanded the total area of stone-paved streets at Pom- iron deposits. The concentration of iron materials in peii. Other pavements from this period were repairs these locations suggests that these repairs were a tem- to the Augustan, and earlier, lava-stone surfaces, and porary, emergency procedure for streets that were too although repairs were necessary, they were also con- important for the city’s post-earthquake recovery to tinuous and disruptive. On average Pompeii witnessed be ripped up for repaving in stone. one or two paving events every year that interrupted The decision to repair these most-used streets with traffic for weeks, while the transportation of materi- iron materials and not to refit them with stone as als for these repairs undoubtedly added to the traffic was being done at the same time in other, less crucial (and the ruts) on streets not under repair. By 79 C.E., streets indicates that the most important factor in de- a third of the city streets were still made of beaten ash. veloping the use of iron materials was time. Consider During the nearly two decades between the the deeply rutted northern half of Via Stabiana be- earthquake(s) and the eruption, street repairs were tween Via dell’Abbondanza in the south and Via della secondary to rebuilding the city’s architecture. Only Fortuna/Via di Nola in the north. Assuming that the 12 sections of street were repaired during this time, stone masons could cut, shape, and closely fit two or and these likely clustered in the 70s rather than the 60s three rows of stones across the street each day (i.e., an C.E. Evidence for the delay in street repairs is found area 5 Roman feet long by ca. 13 Roman feet wide), it in what was occurring on the day of the eruption, would require 169 days (almost half a year) to repave when four paving projects were underway simultane- the entire section if they worked from one end to the ously. Three projects were repairs in important east– other, or 84 days if teams worked from both ends in- west thoroughfares across the southeast of the city,102 wards simultaneously, or 42 days if additional teams while the fourth was paving a street in the northwest, worked outwards from the middle as well. By con- almost certainly for the first time.103 This means that, trast, one gang making 10 deposits a day (one hour on average, paving in 79 C.E. was almost six times per pour, 10 pours per day) could complete the 154

101 Poehler 2017, 54. Our chronology of paving is based on Poehler and Crowther 2018. 102 Via di Castricio and Vicolo delle Conciapelle. 104 Poehler and Crowther 2018, table 1. 103 Vicolo del Fauno. 105 Poehler 2016, 186–91, gs. 6.10–13.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . iron repairs106 in this same section in just less than a lets was 4.4 sestertii/kg.113 To put that price into the third of the time estimated for the fastest repaving with Pompeian economic context, such a billet (6.65 kg, stone.107 If we imagine two gangs each making 12 pours valued at 29 sestertii) weighed the same as a modus a day, the repairs would be complete in less than a of wheat, which would have fed a man for a week but week. Most importantly, this iron repair work could be which cost a quarter as much (7 sestertii). To a work- done without ever having to do more than stall traffic ing Roman, a billet was worth about a week of wages.114 on those streets. These estimates make the iron repairs Translated into iron poured (or pounded) into the between 2.7 and 26.4 times more efficient in terms of streets, that same billet could have been sufficient to repair time. When one factors in the difference be- make no more than four small plugs 6 cm in diameter tween causing a momentary delay versus a weeks-long and 7 cm deep (about the size of a teacupful of iron). detour, the savings in repair time is enormous. Many examples, however, show large areas of staining Perhaps the weightiest reason for the iron repairs that suggest the entire contents of a crucible. For this was their relative cost. Finished raw iron is cheaper estimate, we will assume that an average repair used and stone pavements are more expensive than one one billet. In this section of Via Stabiana, then, the might expect. Evidence from inscriptions provides a iron materials for the 154 repairs (excluding droplets good baseline for the cost of stone pavements, which and splatters) would have cost a minimum of 4,460 Duncan-Jones has found to average 22.5 sestertii per sestertii. For all the iron repairs we have documented, Roman foot.108 However, because this average is de- the cost would have been at least 8,960 sestertii. The rived from extramural streets and does not consider cost of fuel (charcoal) has not been figured in here but the varying widths of urban streets, we have created a can be considered negligible. No price for charcoal has formula to calculate the real costs of paving based on been recorded, but estimates have assumed it to be of area.109 For example, we estimate that the northern sec- very low value.115 On the other hand, if we assume that tion of Via Stabiana described above cost nearly 14,000 the workers were paid (not the only option) and apply sestertii,110 which is nearly three times what the public the time estimated above, then employing 15 men paid in the same period for the monumental basin in five men to run the furnaces and five in each of two the Forum Baths.111 Furthermore, this deeply rutted gangs to pourfor 10 days would add another 2,000 pavement on Via Stabiana was the replacement of an sestertii to the cost of the iron used for Via Stabiana. earlier lava-stone surface, and in 79 C.E. it was in need Still, the total cost (6,460 sestertii) is less than half the of paving for a third time. In fact, across the entire city cost of a new stone pavement. the total cost of repaving (ca. 250,000 sestertii) was The estimate of the volume of material used, how- approaching the cost of the initial pavings (360,000 ever, does not fully take into account the many repairs sestertii).112 In view of the costs in time and money, in the street where ruts and holes were filled by an the desire to find an alternate solution must have been iron-rich slurry of slag, terracotta, ceramics, and stone. strong indeed. Such repairs demonstrate not only that there was much To an average Pompeian, however, iron was ex- more material poured into the streets at Pompeii than pensive, too. Closely contemporary documents from is included in our estimate but also that there are other Vindolanda indicate that the price of refined iron bil- materials with different costs to consider. Unlike iron, there is no known valuation for slag, but there is good reason to believe it was cheap, even free, while still 106 is number includes solid iron as well as staining and ex- cludes droplets (24) and splaers (15). 107 Craddock (1995, 245–46) remarks that a single 19th-century Indian furnace could yield three blooms in a working day, 113 Bray 2010. We may wonder if this price is articially low which seems to have been the norm. If a crucible furnace were because of the proximity of Vindolanda to iron sources and the only twice as ecient, Pompeii would have needed only two or purchasing power of the army. As it relates to our purposes, how- three to support the iron repairs project. ever, it might not maer, as Pompeii, too, was well connected by 108 Duncan-Jones 1982, 124–25. sea to the iron-producing areas, and if bought by the municipal- 109 Poehler and Crowther 2018, 599–600. ity, the city would have had its own leverage in negotiations. 110 On Via Stabiana, see Poehler and Crowther 2018, online 114 Hawkins 2016, 169 n. 118. Supplementary Table 1: PE_006. 115 Landels 2000, 31–3. Carrier (2017, 233 n. 757), however, 111 CIL 10 817. believes Landels underestimates the price of charcoal, especial- 112 Poehler and Crowther 2018, 600. ly in Alexandria.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms being a desirable commodity for traders to transport.116 Indeed, bricks have been offered as the ballast for the Much usable iron was still encased within the slag of return trips in the iron ore trade.124 Of course, iron the original ore.117 Romans knew about these remnant itself was a commodity shipped across the Mediter- metals; slag from gold production was reprocessed,118 ranean both as ore and as finished iron bars.125 In fact, slag collectors (scaurarii) were active at Roman mines the eleven ships wrecked off Saintes-Maries-de-la-Mer in Iberia,119 and metallurgical research on stratified slag (dating from the first century B.C.E. to the first cen- heaps has sometimes found the latest materials to be tury C.E.) carried 500 tons of iron bars among them.126 the lowest in iron content, suggesting they might have For evidence closer to Pompeii, there is the report of been reprocessed.120 In addition, as a commodity, slag Diodorus Siculus that raw iron blooms were brought was not subject to import and export taxes. The Lex from Populonia in Tuscany to Puteoli for refining and Portorii Asiae, reinscribed in 62 C.E., reads: “Whatever for trading elsewhere.127 This Elban ore is likely where is no longer useful, in respect to stone or rock, what- Pompeii got its iron, at least until the end of the first ever has been mined in order to extract gold, silver, century B.C.E.128 After that time, as one of the largest [copper?], iron, lead, orichalcum, or whatever anyone and most long-lived sites in the Mediterranean, Popu- may carry of these, he is not to pay customs.”121 This lonia became a prolific source of slag. It is estimated clause suggests that ships departing from metalwork- that between two and three million tons of slag were ing areas could make a commodity of their ballast and produced in antiquity, with as much as 80% of that bring slags to places like Pompeii where they might having been reprocessed in the early 20th century.129 be in demand. With mounds of slag available for collection and free Although slag itself has not been found in ship- of customs duties, it is possible that merchants sail- wrecks, other evidence from wrecks has demonstrated ing south from Tuscany, even if they already carried that Romans were entrepreneurial in their approach to blooms of iron, might want to make use of and com- finding salable materials for ballast and for dunnage modify this material. Once in the Bay of Naples, these (packing material), just as early modern sailors were.122 ships would be emptied of both cargo and ballast and Millstones and querns, building materials including would be in need of both for their next voyage. One brick, tile, timber, and blocks of stone, other raw mate- of the wrecked ships excavated at San Rossore, Pisa, rials such as coal and glass, and even foodstuffs have all though half a century earlier than the ironwork at Pom- been named as ballast or dunnage in ancient cargoes.123 peii, offers a potential clue about the full cycle of this exchange. Ship B was laden with amphoras containing wine, fruit, nuts, and olives that were packed using 116 Van Oss (2014) found slag sales in 2013 to have averaged small stones of lava, tuff, and lapilli as dunnage. These $17/ton. Van Oss (2002) found prices ranged from $3/ton materials, along with amphoras filled with sand also for domestic air-cooled slag to $35/ton for imported slags and from Ship B, are all sourced to the Vesuvian region.130 $57/ton for granulated slag. With iron and iron slag sources nearby at Elba and on 117 Supra n. 4. the mainland, perhaps ships trading up and down the 118 Plin., HN 33.69. 119 Pérez Macias et al. 2014, 19–22. Italian peninsula brought wine up from the Adriatic 120 In this argument, the slag heap represents the discard from a second smelting, which buried the higher iron content slag from earlier activities. Forbes (1950, 396–97), however, sug- 187). Coal: Smith 1997, 319. Glass: Stern 1999, 475. gests “this should be conrmed by further analyses.” 124 Darvill and McWhirr 1984, 250. 121 Translated in Bernard 2014, 182. 125 Parker (1992, 185, 381) reports a mass 25 m x 15 m of 122 On protable ballast, see Gill 1988, 2. Martin Frobish- high-quality iron ore in the Fuenterrabia wreck and slag as a pos- er’s arctic voyage in 1577 carried as ballast 500 Russian “iron sible cargo in the San Vicenzo A wreck. Taylor (1998, 117) and stones” that could be used for barter or as an emergency source Maingly et al. (2001, 80) argue for iron ore as “saleable ballast.” of iron (McGhee 2001, 169–71). Perhaps the strangest form of 126 Pagès et al. 2008; Pagès et al. 2011. ballast was that of animal mummies exported from Egypt in the 127 Diod. Sic. 5.13.1–2. 19th century (Morgan and McGovern-Human 2008, 585). 128 Saredo Parodi 2013, 25. 123 Millstones: Buckland and Sadler 1990, 116; Alfonsi and 129 Saredo Parodi 2013, 18–22. Gandolfo 1997, 69; Sidebotham 2008, 313 n. 30. Querns: Al- 130 On Ship B, see Bruni (2000, 42–5) for an overview, Mat- len and Fulford 1999, 176. Documented and expected building tioli et al. (2000, 131–41) on the amphoras, and Giachi and materials include vaulting tubes (Vann 1993, 34), bricks (Tch- Pallecchi (2000, 350) and Pecchioni et al. (2007, 18–9) on the ernia 2011, 159 n. 12; Bang 2016, 92), and tiles (Parker 2008, dunnage materials.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . coast packed in Vesuvian materials131 and later returned John Leyland, writing in 1546, remarked that the south balanced by Tuscan iron and slag. It appears that volume of Icelandic cobbles from the cod trade was these traders had opportunistically commoditized sufficient “to pave the whole of Kingston from one their dunnage using building materials from near Na- end to the other.”134 In 18th-century New York City, ples for sale farther north, and it seems unlikely that streets were paved in flat pieces of Belgian granite, they would have neglected to monetize their ballast and during World War II, rubble from the destruction for return trips south. of Bristol crossed the Atlantic as ballast and served as the foundation for New York’s FDR Drive.135 Slag was also a common ballast material in the early modern There is still much to do on this topic at Pompeii, period; San Juan, Puerto Rico, has a long history of especially surveying the remaining streets and chemi- paving its streets in blocks cut from European furnace cally testing the iron and slags. Researchers must also slag.136 Such instances of ballast being repurposed for look for this new paving technology beyond Pompeii. construction material,137 for street surfaces, and even Like the drainage and traffic systems identified at for paving streets in reused slag are historical parallels Pompeii,132 this infrastructural technology was prob- for Pompeii. Like the transatlantic trade of the early ably not unique to that city. Indeed, we have noted one modern period, the boom in Mediterranean shipping example of Type 2 (solid with staining) at Paestum, at in the early imperial period not only increased the ac- the junction of the cardo and decumanus. The pres- tual amount of ballast materials138 but also made them ent results, however, do have potentially important more noticeable. Surely, their conspicuous presence implications. Beyond documenting a novel method made experimentation likely. of street repair, our research disproves the lingering Indeed, related experimentation was already taking misconception among nonspecialists that Romans did place in Roman construction. It is difficult to separate not, or could not, melt iron. Our results represent the the contemporary revolution in the Roman use of first large-scale attestation of the Roman use of mol- concrete from the idea at Pompeii of repairing streets ten iron and shows that we might have misunderstood with another liquefied material. As the Domus Aurea the Romans’ relationship to this technology. Evidence and the Domus Flavia demonstrate, by the 70’s C.E. for molten iron is infrequent not because the Romans Roman architects had become expert in designing and lacked the technology to produce it, but because they building structures of poured concrete. In the Pom- lacked appropriate applications for it. It was not a goal peian context, filling ruts and the voids below paving they were unable to achieve, it was an opportunity they stones was another experiment in the use of liquefied did not know they desired. At Pompeii, we now have solids, one in which the force of adhesion was added to one such opportunity and application. compression. In fact, remnants of opus signinum found Although the process of remelting and pouring adhering to the inside of ruts on Via di Mercurio sug- out ferric materials onto dilapidated thoroughfares gest earlier attempts to apply similar concepts.139 In was probably rare in antiquity, Pompeii would not be alone in reusing ballast materials to pave their streets.133 134 Quoted in Peacock et al. 2007, 28. 135 Burström 2017, 61–7. 131 Volcanic materials from southern Campania were trans- 136 ese silvery-blue blocks are called adoquines. Pabón- ported around the Mediterranean to be used as building mate- Charneco 2016, 81, g. 5.1. rials and bonding agents, most famously to Caesarea Maritima 137 In the 18th and 19th centuries, buildings in La Rochelle, where more than 17,000 m3 (or 13,000 tons) of pulvis puteola- France, were made of substantial quantities of Canadian granite nus was delivered in at least 45 ships. See Arnaud (2014, 168) as (Peacock et al. 2007, 29). well as Jackson et al. (2014, 154–59) on the sourcing of this ma- 138 e Shipwrecks Database of the Oxford Roman Econ- terial, along with other elements of mortars. Lancaster (2015, omy Project shows the greatest number of shipwrecks oc- 30) notes that Vesuvian scoria were essential in the construction curring in the century between 50 B.C.E. and 50 C.E. hp:// of the Pantheon, the Baths of Trajan, and the Baths of Carracalla, oxrep.classics.ox.ac.uk/databases/shipwrecks_database/. and that Sardinian scoriae similarly piggy-backed on the trade in 139 Especially between the middle stepping-stones on Via di millstones with Carthage. Marra et al. (2013) have chemically Mercurio between VI 8, 22, and VI 10, 6. Why Pompeians did traced construction materials at Rome to the Pompeian region. not use cement for their street repairs is unknown, though it was 132 Poehler 2012, 2017. likely a combination of factors including (1) the duration of cur- 133 Burström 2017, 51–9. ing cement compared to iron or even stone repairs; (2) the high

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms Early Imperial architecture, stonecutters were increas- Benton, J. 2014. “Bakers and Bakeries in the Roman Empire: ingly replaced by carpenters and bricklayers. At Pom- Production, Power, and Prestige.” Ph.D. diss., University peii, it seems silicarii were sometimes supplanted by of Virginia. Benvenuti, M., I. Mascaro, P. Costagliola, G. Tanelli, and A. charcoal makers and blacksmiths who discovered that Romualdi. 2000. “Iron, Copper and Tin at Baratti (Popu- they could pour molten iron material into inaccessible lonia): Smelting Processes and Metal Provenances.” His- spaces, binding the paving stones from below and fus- torical Metallurgy 34(2):67–76. ing together the iron streets of Pompeii. Bernard, S. 2014. “Ballast, Mining, and Stone Cargoes in the Lex portorii Asiae.” ZPE 191:182–84. Eric E. Poehler Bray, L. 2010. “‘Horrible, Speculative, Nasty, Dangerous’: As- Classics Department sessing the Value of Roman Iron.” Britannia 41:175–85. 527 Herter Hall Bruni, S. 2000. “Il porto urbano di Pisae ei relitti del comp- University of Massachusetts Amherst lesso ferroviario di ’Pisa S. Rossore.’” In Le navi antiche di Amherst, Massachusetts 01003 Pisa: Ad un anno dall’inizio delle ricerche, edited by S. Bruni, [email protected] 21–79. Florence: Edizioni Polistampa. Burström, M. 2017. Ballast: Laden with History. Lund: Nor- Juliana van Roggen dic Academic Press. 1642 NE 3rd Avenue Buckland, P.C., and J. Sadler. 1990. ”Ballast and Building Camas, Washington 98607 Stone: A Discussion.” In Stone: Quarrying and Building in [email protected] England AD 43–1525, edited by D. Parson, 114–25. Chich- ester, U.K.: Phillimore. Benjamin M. Crowther Carrier, R. 2017. The Scientist in the Early Roman Empire. Classics Department Durham, N.C.: Pitchstone. University of Texas at Austin Cech, B., and T. Rehren, eds. 2014. Early Iron in Europe. Mon- 2210 Speedway, Stop C3400 tagnac, France: Éditions M. Mergoil. Austin, Texas 78712 Cleere, H. 1958. “Roman Domestic Ironwork as Illustrated [email protected] by the Brading, Isle of Wight, Villa.” BIALond 1:55–74. ———. 1971. “Ironmaking in a Roman Furnace.” Britan- nia 2:203–17. ———. 1976. “Some Operating Parameters for Roman Iron- Works Cited works.” BIALond 13:233–46. Aitchison, L. 1960. A History of Metals. Vol. 1. New York: Cleere, H., and D. Crossley. 1985. The Iron Industry of the Interscience. Weald. Leicester: Leicester University Press. Alfonsi, H., and P. Gandolfo. 1997. “L’épave Sanguinaires A.” Coghlan, H.H. 1956. Notes on Prehistoric and Early Iron in the CahArchSubaq 13:35–74 Old World. Oxford: Oxford University Press. Allen, J.R.L., and M. Fulford. 1999. “Fort Building and Mil- ———. 1977. Notes on Prehistoric and Early Iron in the Old itary Supply Along Britain’s Eastern Channel and North World: Including a Metallographic and Metallurgical Exami- Sea Coasts: The Later Second and Third Centuries.” Bri- nation of Specimens Selected by the Pitt River Museum. 2nd tannia 30:163–84. ed. Oxford: Oxford University Press. Arnaud, P. 2014. “Maritime Infrastructure: Between Public Collingwood, R.G., and J.N.L. Myres. 1936. Roman Britain and Private Initiative.” In Infrastruktur und Herrschafts- and the English Settlements. Oxford: Clarendon. organisation im Imperium Romanum, edited by A. Kolb, Craddock, P. 1995. Early Metal Mining and Production. Wash- 161–79. Berlin: De Gruyter. ington, D.C.: Smithsonian Institution Press. Bang, P. 2016. “Beyond CapitalismConceptualising An- ———. 2003. “Cast Iron, Fined Iron, Crucible Steel: Liquid cient Trade Through Friction, World Historical Context Iron in the Ancient World.” In Mining and Metal Produc- and Bazaars.” In Dynamics of Production in the Ancient Near tion Through the Ages, edited by P. Craddock and J. Lang, East: 1300–500 BC, edited by J.C. Moreno García, 91–108. 231–57. London: British Museum Press. Oxford: Oxbow. ———. 2008. “Mining and Metallurgy.” In The Oxford Hand- Bayley, J., and T. Rehren. 2007. “Towards a Functional and book of Engineering and Technology in the Classical World, Typological Classification of Crucibles.” In Metals and edited by J.P. Oleson, 93–120. Oxford: Oxford Univer- Mines: Studies in Archaeometallurgy, edited by S. La Niece, sity Press. D. Hook, and P. Craddock, 46–55. London: Archetype. ———. 2009. Scientific Investigation of Copies, Fakes and Forg- eries. London: Butterworth-Heinemann. Crew, P. 2013. “Twenty-Five Years of Bloomery Experiments: Perspectives and Prospects.” In Accidental and Experimen- viscosity of cement, which would not allow it to penetrate small tal Archaeometallurgy, edited by D. Dungworth and R. cracks in the pavement; and (3) the growing Roman awareness Doonan, 25–50. London: Historical Metallurgy Society. of cement’s lesser tensile strength and concern over its durabil- Crew, P., M. Charlton, P. Dillmann, P. Fluzin, C. Salter, and ity in this context.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . E. Truffant. 2011. “Cast Iron from a Bloomery Furnace.” In The Excavations at Hengistbury Head, Hampshire in In The Archaeometallurgy of Iron: Recent Developments 1911–12, edited by J.P. Bushe-Fox, 72–83. London: So- and Scientific Research, edited by H. Jiří, H. Cleere, and ciety of Antiquaries. M. L’ubomír, 239–62. Prague: Institute of Archaeology Gralfs, B. 1988. Metallverarbeitende Produktionsstätten in Pom- of the ASCR. peji. Oxford: British Archaeological Reports. Curle, J. 1911. A Roman Frontier Post and Its People: The Fort Hawkins, C. 2016. Roman Artisans and the Urban Economy. of Newstead in the Parish of Melrose. Glasgow: J. Macle- Cambridge: Cambridge University Press. hose and Sons. Humphrey, J.W., J.P. Oleson, and A.N. Sherwood. 1998. Greek Curtis, R.I. 1988. “A. Umbricius Scaurus of Pompeii.” In and Roman Technology: A Sourcebook. London: Routledge. Studia Pompeiana and Classica in Honor of Wilhelmina F. Humphris, J., and C. Carey. 2016. “New Methods for Inves- Jashemski, Vol. 1, edited by R.I. Curtis, 19–49. New Ro- tigating Slag Heaps: Integrating Geoprospection, Exca- chelle, N.Y.: Caratzas. vation and Quantitative Methods at Meroe, Sudan.” JAS Darvill, T., and A. McWhirr. 1984. “Brick and Tile Produc- 70:132–44. tion in Roman Britain: Models of Economic Organisa- Jackson, M., G. Vola, E. Gotti, and B. Zanga. 2014. “Roman tion.” WorldArch 15(3):239–61. Sea-Water Concretes and their Material Characteristics.” Davies, O. 1935. Roman Mines in Europe. Oxford: Claren- In Building for Eternity: the History and Technology of Ro- don Press. man Concrete Engineering in the Sea, edited by C. Bran- Deglatigny, L. 1931. Inventaire archéologique de la Seine Inféri- don, R. Hohlfelder, M. Jackson, and J. Oleson, 141–87. eure. Évreux, France: Hérissey. Oxford: Oxbow. Dicus, K. 2014. “Resurrecting Refuse at Pompeii: The Use- Kelleher, B.D. 1982. The Vatican Collections: The Papacy and Value of Urban Refuse and Its Implications for Interpreting Art. New York: Abrams. Archaeological Assemblages.” In TRAC 2013: Proceedings Lancaster, L. 2015. Innovative Vaulting in the Architecture of the of the Twenty-Third Annual Theoretical Roman Archaeology Roman Empire, 1st–4th Centuries CE. Cambridge: Cam- Conference, edited by H. Platts, J. Pearce, C. Barron, J. Lun- bridge University Press. dock, and J. Yoo, 56–69. Oxford: Oxbow. Lang, J. 2017. “Roman Iron and Steel: A Review.” Materials Duncan-Jones, R. 1982. The Economy of the Roman Empire: and Manufacturing Processes 32(7–8):857–66. Quantitative Studies. New York: Cambridge University Landels, J. 2000. Engineering in the Ancient World. 2nd ed. Press. Berkeley: University of California Press. Ellis, S. 2011. “The Rise and Re-organization of the Pom- Margary, I.D. 1965. Roman Ways in the Weald. London: peian Salted Fish Industry.” In The Making of Pompeii: Phoenix House. Studies in the History and Urban Development of an Ancient Marra, F., E. D’Ambrosio, G. Sottili, and G. Ventura. 2013. Town, edited by S. Ellis, 59–88. Portsmouth, R.I.: Journal “Geochemical Fingerprints of Volcanic Materials: Identi- of Roman Archaeology. fication of a Pumice Trade Route from Pompeii to Rome.” Ellis, S., A. Emmerson, K. Dicus, G. Tibbott, and A. Pavlick. Geological Society of America Bulletin 125(3–4):556–77. 2015. “The 2012 Field Season at I.1.1–10, Pompeii: Pre- Mattingly, D.J., D. Stone, L. Stirling, and N. Ben Lazreg. 2001. liminary Report on the Excavations.” Fasti Online: Docu- “Leptiminus (Tunisia): A ‘Producer’ City?” In Economies ments & Research 328:1–21. Beyond Agriculture in the Classical World, edited by D.J. Fieldhouse, W.J., T. May, and F.C. Wellstood. 1931. The Mattingly, 66–89. London: Routledge. Romano-British Industrial Settlement near Tiddington Strat- Mattioli, S., S. Mazzocchin, and M. Pavoni. 2000. “Anfore del- ford-upon-Avon. Birmingham, U.K.: W.I. Rodway. la nave B.” In Le navi antiche di Pisa: Ad un anno dall’inizio Fluzin, P. 2000. “Ponte di Val Gabbia III (Bienno): Les pre- delle ricerche, edited by S. Bruni, 131–47. Florence: Ed- miers résultats des études métallographiques.” In Il ferro izioni Polistampa. nei Alpi: Giacimenti, miniere, e metallurgica dall’antichità al May, T. 1904. Warrington’s Roman Remains: The Roman Forti- XVI secolo, edited by C. Cucini Tizzoni and M. Tizzoni, fications, Potters’ Kilns, Iron and Glass Furnaces, and Bronze 24–31. Bienno: Comune di Bienno. Founders’ and Enamellers’ Workshop, Discovered at Wilder- Forbes, R.J. 1950. Metallurgy in Antiquity: A Notebook for Ar- spool and Stockton Heath, near Warrington. Warrington, chaeologists and Technologists. Leiden: Brill. U.K.: Mackie. ———. 1966. Studies in Ancient Technology. 2nd ed. Vol. 6. McGhee, R. 2001. Arctic Voyages of Martin Frobisher. Mon- Leiden: Brill. treal: McGill-Queen’s University Press. Giachi, G., and P. Pallecchi. 2000. “Analisi preliminari sui Meier-Arend, W. 1984. “Katalog der Metallfunde: Ein Ver- materiali.” In Le navi antiche di Pisa: Ad un anno dall’inizio wahrfund des 4. Jahrhunderts aus dem Königsforst bei delle ricerche, edited by B. Bruni, 348–51. Florence: Ed- Köln.” BJb 184:340–70. izioni Polistampa. Monteix, N. 2016. “Perceptions of Technical Culture Among Gill, D. 1988. “Silver Anchors and Cargoes of Oil: Some Ob- Pompeian Élites, considering the Cupids Frieze of the Casa servations on Phoenician Trade in the Western Mediter- dei Vettii.” In Antike Wirtschaft und ihre kulturelle Prägung ranean.” PBSR 56:1–12. (2000 v. Chr.–500 n. Chr.): The Cultural Shaping of the An- Gowland, W. 1915. “Appendix II: Report on the Metallurgi- cient Economy, edited by K. Droß-Krüpe, S. Föllinger, and cal Remains from the Excavations at Hengistbury Head.” K. Ruffing, 199–221. Wiesbaden: Harrasowitz.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms Morgan, L., and S. McGovern-Huffman. 2008. “Noninvasive ———. 2017. The Traffic Systems of Pompeii. New York: Ox- Radiographic Analysis of an Egyptian Falcon Mummy ford University Press. from the Late Period 664–332 BC.” Journal of Avian Biol- Poehler, E.E., and B.M. Crowther. 2018. “Paving Pom- ogy 39(5):584–87. peii: The Archaeology of Stone-Paved Streets.” AJA Morton, R., and J. Wingrove. 1970. “Reduction of Iron from 122(4):579–609. its Ore in the Medieval Bloomery.” Steel Times 198(4):3–8. Poehler, E.E., and S. Ellis. 2013. “The Pompeii Quadriporti- Pabón-Charneco, A. 2016. The Architecture of San Juan de cus Project: The Eastern Side and Colonnade.” Fasti On- Puerto Rico: Five Centuries of Urban and Architectural Ex- line: Documents & Research 249:1–14. perimentation. London: Routledge. Rackham, O. 2001. Trees and Woodland in the British Land- Pagès, G., P. Dillmann, P. Fluzin, and L. Long. 2011. “A Study scape: The Complete History of Britain’s Trees, Woods and of the Roman Iron Bars of Saintes-Maries-de-la-Mer Hedgerows. London: Phoenix. (Bouches-du-Rhône, France): A Proposal for a Compre- Read, T. 1934. “The Early Casting of Iron: A Stage in Iron hensive Metallographic Approach.” JAS 38:1234–252. Age Civilization.” Geographical Review 24(4):544–54. Pagès, G., L. Long, P. Fluzin, and P. Dillmann. 2008. “Réseaux Rehder, J.E. 2000. The Mastery and Uses of Fire in Antiquity. de production et standards de commercialisation du fer Montreal: McGill-Queen’s University Press. antique en Méditerranée : Les demi-produits des épaves Rihll, T.E. 2013. Technology and Society in the Ancient Greek romaines des Saintes-Maries-de-la-Mer (Bouches-du- and Roman Worlds. Washington, D.C.: American Histori- Rhône).” Revue archéologique de Narbonnaise 41:261–83. cal Association, Society for the History of Technology. Parker, A. 1992. Ancient Shipwrecks of the Mediterranean Rispoli, P., and R. Paone. 2011. “Pompeii scavi, canale con- and the Roman Provinces. BAR-IS 580. Oxford: Tempus te di Sarno: Lavori di sistemazione e rifunzionalizzazione Reparatum. 2009–2011.” RStPomp 22:126–33. ———. 2008. “Distributions and Wreck Locations: The Ar- Russell, B. 2011. “Lapis transmarinus: Stone-Carrying Ships chaeology of Roman Commerce.” In The Maritime World and the Maritime Distribution of Stone in the Roman Em- of Ancient Rome, edited by R. Hohlfelder, 177–96. Ann pire.” In Maritime Archaeology and Ancient Trade in the Med- Arbor: University of Michigan Press. iterranean, edited by D.J. Robinson and A. Wilson, 137–52. Peacock, D., D. Williams, and S. James. 2007. “Basalt as Oxford: Oxford Centre for Maritime Archaeology. Ships’ Ballast and the Roman Incense Trade.” In Food for Russell, M. 2003. Piltdown Man: the Secret Life of Charles Daw- the Gods, edited by D. Peacock and D. Williams, 28–70. son and the World’s Greatest Archaeological Hoax. Stroud, Oxford: Oxbow. U.K.: Tempus. Pecchioni, E., E. Cantisani, P. Pallecchi, F. Fratini, A. Buccian- Saredo Parodi, N. 2013. Populonia: Inferno o paradiso? Il polo ti, E. Pandeli, and S. Conticelli. 2007. “Characterization of siderurgico di Populonia nell’antichità, un tentativo di quan- the Amphorae, Stone Ballast and Stowage Materials of the tificazione. Rome: Aracne. Ships from the Archaeological Site of Pisa-San Rossore, Schubert, H.R., and A.H. Ingen Housz. 1957. History of the Italy: Inferences on their Provenance and Possible Trad- British Iron and Steel Industry from ca. 450 B.C. to A.D. 1775. ing Routes.” Archaeometry 49(1):1–22. London: Routledge & Kegan Paul. Peña, J.T. 2007. Pottery in the Archaeological Record. Cam- Sidebotham, S.E. 2008. “Archaeological Evidence for Ships bridge: Cambridge University Press. and Harbor Facilities at Berenike (Red Sea Coast), Egypt.” Peña, J.T., and M. McCallum. 2009a. “The Production and In The Maritime World of Ancient Rome, edited by R. Hohl- Distribution of Pottery at Pompeii: A Review of the Evi- felder, 305–24. Ann Arbor: University of Michigan Press. dence; Part 1, Production.” AJA 113(1):57–79. Sim, D. 2011. Roman Iron Industry in Britain. New York: His- ———. 2009b. “The Production and Distribution of Pot- tory Press. tery at Pompeii: A Review of the Evidence; Part 2, The Smith, A.H.V. 1997. “Provenance of Coals from Roman Sites Material Basis for Production and Distribution.” AJA in England and Wales.” Britannia 28:297–324. 113(2):165–201. Sogliano, A. 1899. “Pompei.” NSc 24:439–448. Pérez Macias, J.A., A. Martins, and J.X. de Matos. 2014. “The ———. 1900. “Pompei.” NSc 25:594–603. Scavrarii of Vipsaca.” In I codici minerari nell’Europa prein- Stern, E.M. 1999. “Roman Glassblowing in a Cultural Con- dustriale: Archeologia e storia, edited by R. Farinelli and G. text.” AJA 103(3):441–84. Santinucci, 15–22. Florence: Insegna del Giglio. Stewart, N.S. 2002. “The Technology and Control of Min- Pleiner, R. 2000. Iron in Archaeology: The European Bloomery ing in Roman Britain.” Ph.D. diss., University of Oxford. Smelters. Prague: Archeolgický ústav AV CR. Straker, E., and I. Margary. 1938. “Ironworks and Commu- ———. 2006. Iron in Archaeology: Early European Black- nications in the Weald in Roman Times.” Geographical smiths. Prague: Archeolgický ústav AV CR. Journal 92(1):55–60. Poehler, E.E. 2012. “The Drainage System at Pompeii: Mech- Taylor, J. 1998. “Late Iron Age Ballast-Quarries at Hengist- anisms, Operation and Design.” JRA 25:95–120. bury Head, Dorset.” OJA 17(1):113–19. ———. 2016. “Measuring the Movement Economy: A Net- Tchernia, A. 2011. Les Romains et le commerce. Naples: Cen- work Analysis of Pompeii.” In The Economy of Pompeii, ed- tre Jean Bérard. ited by M. Flohr and A. Wilson, 163–208. Oxford: Oxford Tholander, E. 1987. Experimental Studies in Early Iron- University Press. Making. Stockholm: Royal Institute of Technology, De-

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms . ., partment of Metallurgy. Wertime, T. 1961. The Coming of the Ages of Steel. Leiden: Tholander, E., and S. Blomgren. 1985. “On the Classifica- Brill. tion of Ancient Slags by Microstructure Examination.” Iskos 5:415–25. Tsujimura, S. 1991. “Ruts in Pompeii: The Traffic System in the Roman City.” Opuscula Pompeiana 1:58–86. Tylecote, R.F. 1986. The Prehistory of Metallurgy in the Brit- ish Isles. London: Institute of Metals. ———. 1987. The Early History of Metallurgy in Europe. Lon- don: Longman. van Oss, H.G. 2002. “SlagIron and Steel.” U.S. Geological Survey Minerals Yearbook 2002. United States Geologi- cal Survey, Mineral Resources Program. https: // minerals. usgs.gov/minerals/pubs/commodity/iron_&_steel_ slag/islagmyb02.pdf. ———. 2014. “Iron and Steel Slag.” U.S. Geological Survey, Mineral Commodity Summaries, February 2014. United States Geological Survey, Mineral Resources Program. https: // minerals.usgs.gov/minerals/pubs/commodity/ iron_&_steel_slag/mcs-2014–fesla.pdf Vann, R. 1993. “Vaulting Tubes from Caesarea Maritima.” IEJ 43(1):29–34. Varone, A. 1998. “Un pittore al lavoro a Pompei.” In Romana picture: La pittura romana dalle origini all’età bizantina, edited by A. Donati, 302–4. Milan: Electa.

This content downloaded from 130.237.165.40 on Wed, 30 Sep 2020 17:32:02 UTC All use subject to https://about.jstor.org/terms